A phonetic writing system. One letter, one sound, always.
Détournement of the Latin-Greek-Cyrillic tradition: the existing letterforms are rerouted from their conventional orthographic roles into a strictly phonetic system where spelling convention cannot hide pronunciation. Most letters are inherited — the forms that already proved themselves across centuries and billions of hands. A few letters are new, designed for sounds that no existing single letter covers. The new letters are derived from the geometry of the cube, styled to sit alongside the inherited forms. Numbers are standard decimal. The punctuation is five marks. The whole thing reads, counts, and speaks.
Human speech is a lattice. Every sound has an address in three dimensions:
Consonants: Place — where in the mouth: lips, teeth, ridge, palate, velum, glottis. Front to back. Manner — how the air moves: stopped, forced through a gap, redirected through the nose, flowing around the tongue, gliding. Voicing — whether the vocal cords vibrate: yes or no.
Vowels: Height — jaw position: open, mid, close. Front — tongue position: front, central, back. Rounding — lip shape: rounded or spread.
Every language selects a subset of this lattice. English occupies about thirty positions. Serbo-Croatian about thirty. Zulu more, Hawaiian fewer. The alphabet covers the lattice. A speaker of any language picks the letters they need.
Seven vowels, derived from the Serbo-Croatian baseline — the cleanest phonetic system in the European tradition. These seven cover the vowel space without gaps and without redundancy.
Letter Sound Mouth position As in A /a/ open, central father E /e/ mid-front bed, café I /i/ close-front see O /o/ mid-back, rounded so, more U /u/ close-back, rounded moon R /r̩/ syllabic r Serb. prst, Eng. butter Y /j/~/y/ front glide / rounded yes, Fr. tu
A, I, U are the three corners of the vowel triangle — universal across all human languages. E and O fill the mid positions that most European languages distinguish. R is the syllabic r that Slavic languages use as a vowel (prst, krk, trg — words with no other vowel). Y sits on the vowel-consonant boundary, a glide that can function as either.
Five of the seven are the Latin vowels unchanged: A E I O U. R and Y are Latin letters reclassified to acknowledge what the mouth already does with them. No new shapes needed for any vowel.
Consonants are organised by natural groupings. The groupings emerge from the lattice: sounds made at the same place, in the same manner, with the same voicing mechanism cluster together. The alphabet respects these clusters.
At two places of articulation, five sounds form a complete family: a voiceless stop, its voiced partner, a voiceless fricative, its voiced partner, and a nasal. Five sounds, one place, all the major manners represented.
Labial family (lips): P /p/ voiceless stop pen B /b/ voiced stop bed F /f/ voiceless fricative fish V /v/ voiced fricative van M /m/ nasal man
Alveolar family (tongue ridge): T /t/ voiceless stop ten D /d/ voiced stop dog S /s/ voiceless fricative sun Z /z/ voiced fricative zoo N /n/ nasal net
At the velum (back of mouth), the fricative pair is rare in most European languages. What remains is a stop pair and a nasal.
Velar family (back): K /k/ voiceless stop cat G /g/ voiced stop go Ŋ /ŋ/ nasal ring
Ŋ is the first letter that has no single inherited Latin form. It needs a new symbol — one of the cube-derived letters.
Three pairs of sounds that exist only as fricative pairs, without stops or nasals at their place of articulation.
Dental fricatives (tongue between teeth): Θ /θ/ voiceless thin Ð /ð/ voiced the
Postalveolar fricatives (tongue behind ridge): Σ /ʃ/ voiceless ship Ζ̌ /ʒ/ voiced measure
Postalveolar affricates (stop releasing into fricative): Ч /tʃ/ voiceless chin Џ /dʒ/ voiced judge
Θ is borrowed from Greek. Ð is borrowed from Icelandic and Old English, where it served exactly this sound for centuries. The remaining four (Σ for /ʃ/, its voiced partner, and the affricate pair) are new letters derived from the cube geometry. The placeholder symbols above (Σ, Ζ̌, Ч, Џ) indicate the sounds — the actual glyph forms are specified in the scratch variant section.
Note: the Cyrillic letters Ч and Џ are shown as reference points — Cyrillic already solved these sounds. The final shadow forms may draw from them or from the cube, but the phonetic assignment is fixed.
Three consonants that have no voicing partner in common use.
H /h/ glottal fricative hat L /l/ lateral let W /w/ labial-velar glide wet
H is breath without constriction — a singleton because its voiced counterpart (breathy voice) is not phonemic in most languages. L is the lateral — air flows around the tongue's sides. W is a lip-rounded glide, close to U but consonantal.
Three Latin letters are absent because they duplicate sounds that other letters already cover:
C = K before back vowels, S before front vowels. Redundant. Q = K with a U attached by convention. Redundant. X = K + S spoken together. A compound, not a sound.
English words currently spelled with C, Q, or X are respelled using K, S, and KS. Cat becomes KAT. Queen becomes KWIN. Box becomes BOKS. No information is lost. Three letters of confusion are removed.
The letter J is reassigned: where Latin used J ambiguously (English /dʒ/, French /ʒ/, German /j/), the shadow alphabet gives each sound its own letter. J as a symbol is retired; the sounds it represented are handled by Џ, Ζ̌, and Y respectively.
The relationship between voiced and voiceless consonants is the single most important structural pattern in the consonant system. B is P with the vocal cords engaged. D is T with the vocal cords engaged. The mouth position is identical; only the larynx changes.
For the inherited Latin pairs (P/B, T/D, K/G, F/V, S/Z), this relationship is invisible — the letters look nothing alike. A reader must simply memorise that B and P are related. This is a historical accident, not a design choice.
For the new letters, the relationship is made visible. A single modifier mark — the voicing mark — transforms a voiceless letter into its voiced partner. This follows the Japanese dakuten principle (ka か becomes ga が with two strokes) and the Croatian háček principle (S becomes Š, Z becomes Ž, C becomes Č).
Base letter + voicing mark = voiced partner: Θ + mark = Ð (thin → the) Σ + mark = Ζ̌ (ship → measure) Ч + mark = Џ (chin → judge)
In the scratch variant, the voicing mark is a short diagonal stroke crossing the staff at mid-height from left to right — the only mark that crosses the staff body (as opposed to branching from it). It is visually distinct and consistent across all letters it modifies.
This means three base letter forms generate six sounds. The reader who learns the voicing mark learns all six from three shapes plus one principle.
Twenty-six letters. The order follows a principle: vowels first, then consonants grouped by place of articulation, front to back. Within each place, voiceless before voiced, stops before fricatives, nasals last. Singletons at the end.
# Letter Sound Name 1 A /a/ a 2 E /e/ e 3 I /i/ i 4 O /o/ o 5 U /u/ u 6 R /r̩/ er 7 Y /j/ ye 8 P /p/ pe 9 B /b/ be 10 F /f/ fe 11 V /v/ ve 12 M /m/ me 13 T /t/ te 14 D /d/ de 15 S /s/ se 16 Z /z/ ze 17 N /n/ ne 18 K /k/ ke 19 G /g/ ge 20 Ŋ /ŋ/ eng 21 Θ /θ/ theta 22 Ð /ð/ edh 23 Σ /ʃ/ sha 24 Ч /tʃ/ cha 25 H /h/ ha 26 L /l/ el 27 W /w/ wa
Note: Ζ̌ (/ʒ/) and Џ (/dʒ/) are generated from Σ and Ч by the voicing mark and do not need separate positions in the alphabetic order. They exist as sounds but not as independent sort positions, the same way the Japanese voiced kana sort under their unvoiced base forms.
Numbers are written in standard decimal: 0 1 2 3 4 5 6 7 8 9. Arabic-derived positional notation. The system that won. There is no letter-based numeral encoding, no additive system, no shadow mark required to switch modes. Numbers look like numbers. Letters look like letters. No ambiguity.
42 is written 42. 2025 is written 2025.
The ten digit forms in the scratch variant are their standard angular shapes, adapted for straight-line carving the same way the letter forms are. In the pen variant they take their standard handwritten forms. A reader of any Latin-script language already knows them.
The ten decimal digits count along the real number line. But the system described in this document is two-dimensional — the imaginary axis is not a metaphor, it is a structural feature of the phonetic lattice. The imaginary unit i, with the property that i² = -1, is the eleventh necessary number.
The lowercase Latin letter i serves as this numeral. It is not the vowel I (uppercase, third letter of the alphabet, the sound /i/). It is the mathematical constant: the unit rotation that maps the real axis onto the perpendicular. In the shadow alphabet, uppercase I is a sound, lowercase i is a number. The two are visually distinct in both scratch and pen variants — the vowel is a full-height staff, the numeral is a half-height mark with a dot (its standard mathematical typography).
0 1 2 3 4 5 6 7 8 9 i
Eleven symbols. Ten count positions on the line. One rotates off the line into the perpendicular axis where transformations live. Without i, the number system can count letters but cannot express the operations between them — voicing, place-shifting, the algebraic structure that makes the phonetic lattice a lattice rather than a list.
Each letter occupies a position in the alphabetic order (1 through 27). This ordinal position is the letter's real component. But the alphabet also encodes relationships between sounds — voicing pairs, place-of-articulation families, manner groups — that exist on a perpendicular axis to the sequence. These structural relationships are the imaginary component.
In complex number notation, a value z = a + bi has a real part (a, the position on the number line) and an imaginary part (b, the position on the perpendicular axis). The imaginary unit i has the property that i² = -1: the imaginary, operated on itself, flips back into the real but inverted. This is the structure of voicing: a voiceless consonant, passed through the voicing operation, becomes its partner. Applied again, it returns. The operation is its own inverse.
The interval points between integers — the places where one number transitions to the next — are where the imaginary axis intersects. Between P (8) and B (9), the voicing operation lives. Between Θ (21) and Ð (22), the same operation. The real axis counts letters. The imaginary axis encodes the transformations between them.
This is a sketch, not a specification. The decomposition of letter-relationships into real and imaginary components connects the phonetic alphabet to arithmetic in a way that could make both teachable through the other: learn the letters, learn the numbers, and discover that the operations on one mirror the operations on the other. Addition on the real axis is concatenation (spelling). Rotation through the imaginary axis is voicing, or place-shifting, or manner-changing — the phonetic transformations that turn one sound into another.
The full formalisation of this mapping is an open problem. What is clear is that the relationships exist and that complex number notation is the right frame for expressing them, because the structure is genuinely two-dimensional: sequence and transformation are perpendicular axes, and the algebra of each constrains the other.
The imaginary axis operates not only within a single language but across language boundaries. Consider five words from five languages sharing the tel- phoneme: telos (Greek, purpose), tijelo (Croatian, body), htio (Croatian, desired — h silent), tele (Bulgarian, calf), tell (English, signal/narrate). Five unrelated PIE roots (*kwel-, teyh₁-lo-, xъtěti, telh₁-, del-), converging on the same acoustic address across four millennia. The phonetic lattice predicts this: sounds that occupy the same coordinates in the articulatory space — same tongue position, same voicing, same manner — are the same address regardless of which language shipped the packet. The real axis is the sequence within one language. The imaginary axis is the transformation between languages. The tel- cluster is five languages arriving at the same imaginary-axis coordinate from different real-axis positions.
Punctuation is built from short vertical nicks — marks visibly shorter than any letter staff, carrying no branches or voice marks. They are position markers, not phonetic symbols. They instruct the voice: pause here, stop here, ask here, emphasise here, branch here.
In the pen variant, these are ink dots — the nib touches and lifts. In the scratch variant, they are short vertical cuts, about a quarter of staff height. The tool motion is the same straight cut used for everything else in the system. No pits, no stabs, no change in technique.
Five marks, arranged from one to three nicks:
Mark Pattern Function Breath | one nick, centred the brief pause, the comma
Rest : (two nicks stacked) two nicks, vertical the full stop, the period
Rise .' (low + high-right) one nick low, one high-right the question, upward inflection
Strike ! (high + low, gap) one nick high, one low, gap the exclamation, hard stop
Branch :. (two stacked + offset) two nicks stacked, one offset the colon/semicolon, the fork
The accent mark is a short horizontal tick placed above a letter's staff — the only mark that appears above the letter body. It indicates which syllable carries the stress. One accent per word.
record (noun): R E' K O R D (stress on E) record (verb): R E K O' R D (stress on O)
The primary physical form. Designed for carving into wood, bone, stone, and any resistant surface. All strokes are straight. No curves. The same constraints that shaped the Elder Futhark and Ogham.
The twenty-two Latin/Greek letters (A B D E F G H I K L M N O P R S T U V W Y Z, plus Θ) are written in their standard capital forms, adapted for straight-line carving. Where a Latin capital already consists of straight lines (A E F H I K L M N T V W Y Z), it transfers directly. Where it has curves (B D G O P R S U), the curves are straightened into angular approximations — the same adaptation that happened naturally when Latin letters were carved in stone rather than written with a pen.
This is not reinvention. Roman monumental inscriptions already solved this problem two thousand years ago. The scratch variant simply adopts their solutions.
The four new letters (Ŋ, Σ, Ч, and the voicing mark) are built on the staff-and-branch system from the cube geometry:
A vertical staff serves as the spine. Distinguishing features are short diagonal strokes branching rightward from the staff at fixed positions. The position and number of branches encode the letter's identity. This is the Ogham principle: a central axis with lateral marks.
The voicing mark is a short diagonal stroke crossing the staff at mid-height from left to right — the only mark that goes leftward, making it instantly recognisable as a modifier rather than a base letter feature.
The specific branch patterns for the new letters are:
Ŋ (eng, /ŋ/): staff + branch at bottom position Σ (sha, /ʃ/): staff + branch at top position Ч (cha, /tʃ/): staff + branches at top and bottom
Voiced forms (with voicing mark crossing mid-staff): Ð (edh, /ð/): Θ + voicing mark Ζ̌ (zha, /ʒ/): Σ + voicing mark Џ (ja, /dʒ/): Ч + voicing mark
Note: Θ (theta) is inherited from Greek and retains its standard form (circle with a horizontal bar, or in scratch variant: an angular diamond with a cross-bar). The voicing mark transforms it into Ð.
Left to right, as with Latin and Greek.
Inherited letters: standard stroke counts for Latin capitals (1–4 strokes each). New letters: at most 3 strokes (staff + 1–2 branches). New letters with voicing mark: at most 4 strokes. The average letter is 2–3 strokes.
A half-letter-width of empty space between letters. A full letter-width between words. Uniform height across all letters makes it easy to carve in straight lines along any surface.
A cursive adaptation for writing with pen on paper. Inherited letters take their standard handwritten forms. For the cube-derived letters, the straight staff becomes a gentle rightward arc, branches become small loops, and the voicing mark becomes a short cross-stroke. Terminal flourishes distinguish letter groups.
This variant is secondary. It develops naturally once the scratch forms are established, the same way Latin minuscule developed from Roman capitals through centuries of handwriting practice. The details of the pen variant will emerge from use rather than from specification.
English uses all twenty-six base letters plus the three voiced forms generated by the voicing mark. Words are respelled phonetically — one letter per sound, no exceptions.
English word Shadow spelling Sounded out cat K A T k a t dog D O G d o g fish F I Σ f i sh ship Σ I P sh i p thin Θ I N th(voiceless) i n the Ð E th(voiced) e moon M U N m u n ring R I Ŋ r i ng church Ч R Ч ch r ch boy B O Y b o y day D E Y d e y measure M E Ζ̌ R m e zh r judge Џ A Џ j a j queen K W I N k w i n box B O K S b o k s
Serbo-Croatian already has near-perfect phonetic spelling in both Gaj's Latin and Vuk's Cyrillic. The shadow alphabet maps to it directly:
Croatian Sound Shadow A /a/ A B /b/ B C /ts/ T S (compound) Č /tʃ/ Ч Ć /tɕ/ Ч (close enough, or dialect mark) D /d/ D Dž /dʒ/ Џ Đ /dʑ/ Џ (close enough, or dialect mark) E /e/ E F /f/ F G /g/ G H /x/ H I /i/ I J /j/ Y K /k/ K L /l/ L Lj /ʎ/ L Y (compound) M /m/ M N /n/ N Nj /ɲ/ N Y (compound) O /o/ O P /p/ P R /r/ R S /s/ S Š /ʃ/ Σ T /t/ T U /u/ U V /v/ V Z /z/ Z Ž /ʒ/ Ζ̌
The shadow alphabet handles Serbo-Croatian with almost no friction. The mapping is one-to-one for most letters. The compounds (TS for C, LY for Lj, NY for Nj) reflect the actual articulation — these are compound sounds written as compound letters. Honest spelling.
Any language maps to the lattice by selecting its sounds and using the corresponding letters. Languages with sounds not in the base twenty-six (clicks, pharyngeals, tones, retroflex consonants) extend the system using the same cube-derivation method: staff and branches encode identity, voicing mark toggles the larynx. The lattice has addresses for every human sound. The base alphabet covers the most common addresses. Extensions cover the rest.
A short horizontal tick placed above a letter marks the stressed syllable. In the scratch variant this is a short horizontal cut above the staff top. In the pen variant it is a dot above the arc. One accent per word.
record (noun): R E' K O R D (stress on first syllable) record (verb): R E K O' R D (stress on second syllable) permit (noun): P E' R M I T (stress on first syllable) permit (verb): P E R M I' T (stress on second syllable)
The accent solves the problem that English spelling gives no clue about stress. Same letters, different stress, different meaning — the accent makes the distinction visible.
27 letters (7 vowels + 20 consonants) 3 voiced forms (generated by voicing mark, not independent) 1 voicing mark (modifier, not a letter) 11 numerals (0–9 decimal + i imaginary unit) 5 punctuation (nick patterns) 1 accent mark (stress indicator) 1 space (word boundary)
Letters: 27 base + 3 voiced = 30 sounds Structural: 11 numerals + 5 punctuation + 1 accent + 1 space = 18 Total: 48 symbols, of which 23 are inherited Latin/Greek, 4 are cube-derived (3 letters + 1 voicing mark), and 21 are shared with every numerate culture
This section is etymology, not prerequisite. The alphabet can be learned and used without reading it. It explains where the new letters come from and why the system has the structure it does.
A cube has eight corners. Label each corner with a trigram: three binary choices (solid or broken, one or zero) that name the corner's position along three axes. These eight corners are the eight trigrams of the I Ching, the eight vertices of the binary cube.
Trigram Bits Meaning Earth 000 substrate, emptiness Thunder 001 first impulse, nucleation Water 010 deep structure, hidden pattern Lake 011 ambiguity, surface play Fire 100 transformation, clarity Heaven 101 fullness, creative force Wind 110 gentle penetration, subtlety Mountain 111 stillness, completion
The cube has three axes. Each axis connects two opposing faces. When viewed along one axis, the cube casts a shadow — a square silhouette. Three axes, three shadows, three voices in which each corner can speak.
The three voices correspond to the three phonetic axes:
Voice 1: Bonding — place of articulation (lips) Voice 2: Constraint — place of articulation (tongue) Voice 3: Energy — place of articulation (velum)
Eight corners times three voices gives twenty-four positions. But the alphabet has twenty-six to twenty-seven letters because phonetics, not geometry, determines the count. The cube provides the design language — the staff-and-branch construction method — not the size of the inventory.
The original trigram-to-sound mapping (Earth = vowels, Mountain = voiceless stops, Thunder = voiced stops, Wind = voiceless fricatives, Fire = voiced fricatives, Water = nasals, Lake = liquids, Heaven = affricates) remains valid as a mnemonic and as the logic behind the cube-derived letters. But a user does not need to know this mapping to read or write. The letters are just letters. The cube is their history.
The geometry teaches. But the alphabet works without the lesson.